College and Research Libraries Research Notes An Assessment of Choice as a Tool for Selection John P. Schmitt and Stewart Saunders Samuel Johnson noted in 1759 that "a corrupt age has many laws; I know not whether it is equally true, that an ignorant age has many books. " 1 To remedy this sit- uation, book reviewing was established as a service which makes material acquisition more an objective judgment than an intui- tive hunch. Two prospects librarians dread are: (1) being told they do not have ''the definitive work'' in a field and; (2) having a full range of "definitive works" which never circulate. Book reviewing is designed to prevent the first prospect, but little has been done to study the second. This study was designed to examine the relationship between two factors-the strength of a reviewer's recommendation and the subsequent use of that title in a large university library. Is there agree- ment between the reviewer and the stu- dent reader on what constitutes an indis- pensable volume? The review medium selected for the study was Choice, while the library where the materials' use was examined was Pur- due University's General Library, which serves the School of Humanities, Social Sciences, and Education. Choice has estab- lished a unique niche in its service to aca- demic libraries by publishing concise re- views of more than 6,000 books per year. The reviewers are by and large faculty members engaged in undergraduate in- struction who demonstrate an in-depth understanding of a special subject area. The established criterion for inclusion in Choice is "potential use by undergradu- ates. " 2 To be sure, Choice makes no claim as an automatic buying guide for college li- braries, but states as its purpose "to assist the college librarian and his faculty in the selection of current books ... (including) those subject areas which form the basis of the liberal arts curriculum.' ' 3 Part of this assistance to college librari- ans is the assessment of a title in terms of the audience it is most likely to appeal to. Frequently the recommendation is by class rank or status: lower division reader- ship, graduate students and advanced un- dergraduates, faculty and professional au- dience only, etc. Occasionally the recommendation will be sized for an insti- tution: two-year and community college readership, larger research collections, special subject collections, all academic li- braries, etc. The intention is that the re- view's recommendation can thus be scaled to a particular clientele. The ques- tion is raised: how accurate are these rec- ommendations when the circulation rec- ords are examined? It has been shown that Choice, by way of comparison with Library Journal, is likely to review more university press titles, more publications from the social sciences John P. Schmitt is reference librarian in the Purdue University Undergraduate Library, West Lafayette, Indiana, and Stewart Saunders is reference and liaison librarian in the Purdue University Humanities Library. 375 376 College & Research Libraries and humanities (excluding fiction), and is more likely to compare a book to an earlier title or at least discuss its place in the sub- ject literature. 4 Daniel Ream demon- strated that Choice reviewed more titles than three other major review media in 1975, although the ACRL publication took the longest to review new books, a fact that could be attributed to a policy of not reviewing from galley proofs. 5 The scope of reviews to appear in Choice is hinted at in the Bowker Annual where the figures for 1979 indicate that approximately 16 per- cent of all new titles (excluding juvenile books) published that year were reviewed by this periodical. 6 Richard Werking and Charles Getchell have suggested that Choice is a reliable gauge of academic pub- lishing activity by subject area and thus could be manipulated to serve as a book fund allocation device. 7 These studies have demonstrated the unique role Choice has played in assisting academic library collection development, but do not ad- dress the question of how patrons make use of the titles recommended. EXPERIMENTAL PROCEDURE In order to examine the relationship be- tween the strength of the reviews in Choice and the subsequent circulation of the titles reviewed, the authors drew a sample of reviews from Choice, ranked the reviews according to the reviewer's opinion of the book, determined the frequency of circu- lation of each title reviewed, and corre- lated the frequency with the strength of the review. The Purdue General Library provided favorable conditions for the study in that the collections are suffi- ciently large to contain most of the titles reviewed by Choice and in that undergrad- uates account for about 70 percent of the circulation. Because the holdings of the Purdue General Library are primarily in the areas of the humanities and social sci- ences, the authors limited their sample of reviews to these areas. The sampling procedure was designed to meet three objectives: 1. To offer a representative sample of ti- tles reviewed in Choice. September 1983 2. To offer a stratified sample of the hu- manities and the social sciences. 3. To offer a sample of titles which were available for circulation in the Purdue General Library for at least two years but no more than three and one-half years. The objectives were met by selecting a sample of titles which: 1. were reviewed in issues of Choice be- tween November 1978 and April1979; 2. provided a balanced representation of titles in History, Philosophy, English and American Literature, Political Science, So- ciology, and Education; 3. were cataloged by the Purdue Li- braries between Spring 1978 and Summer 1979. Reprints, serials, and works which be- came part of a reference collection or re- serve book collection were eliminated from the sample. The necessity to meet all three objectives thus required a stratified cluster sample rather than a random sample, but the sta- tistical procedures to be used were consid- ered sufficiently robust to accommodate this sampling design. One cause for con- cern arose from the fact that the titles se- lected from each strata of the cluster were not randomly selected but were subject to the chance that they were purchased by Purdue Libraries and cataloged within a certain range of dates. Fortunately, 66 per- cent of the strata was included in the sam- ple, thus minimizing the possibility of dis- tortion due to sampling procedures. The review for each of the titles selected from Choice was ranked according to the strength of the recommendation insofar as it predicted widespread appeal to under- graduates. Titles recommended for an elite or special audience were given a mid- dle ranking. Titles with mediocre or nega- tive reviews were ranked at the bottom as least likely to circulate. The rankings were: 5-Highly recommended for a broad audi- ence. An indispensable volume for all collections; Even the smallest libraries will want to acquire this. 4-Generally recommended for most lev- els. A good piece but not necessarily in- dispensable for everyone; neverthe- less, recommended without hesita- tion. 3-Recommended with limitation. The book is aimed at a specialized audience or a special collection. 2-Reserved recommendation. Some doubts are expressed about the quality, format, or organization of the work. Of interest primarily to large research li- braries with substantial holdings in the area. 1-Not recommended. The circulation record for each title se- lected was examined in the summer of 1981, thus guaranteeing a 2 to 3.5-year test period for each title. Other studies have shown that the circulation record of a title during its first two years on the shelf are a good predictor of future circulation. 8 The number of three-week (student) and two-month (faculty) circulations was recorded for each title. This distinction be- tween users is accurate except in rare in- stances when a faculty member requests a shorter loan period. SPSS programs were used to calculate the relevant statistical tests. RESULTS The circulation pattern of the 310 titles selected for the sample indicates that they are quite typical in their frequency of use. Ninety-four titles (30.3 percent) did not circulate at all during the test period while 114 titles (36.8 percent) circulated one to two times and 102 titles (33 percent) circu- lated three or more times. A dispropor- tionate number of reviews fell in the highly recommended or generally recom- mended categories of ranks 5 and 4. This corresponds with Macleod's findings that few book reviews-about 18 percent- tend not to be positive. 9 A good 188 titles (60.7 percent) were given the green light for college audiences, that is, ranks 5 and 4, while only 122 titles (39.3 percent) were considered either too specialized or inap- propriate for inclusion in a college collec- tion. This imbalance in the distribution of rankings may result from an attempt on the part of the editors of Choice to screen for review those titles most suitable for in- clusion in a college collection. Research Notes 377 A cross-tabulation of circulation with the evaluations of reviewers reveals that the titles with the highest recommenda- tion for undergraduates do indeed circu- late more frequently than do those rated for more specialized audiences. Table 1 shows that only 23 percent of the titles ranked at the top (rank 5) and only 26 per- cent ranked next (rank 4) had never circu- lated. On the other hand 41 percent of the more specialized titles (rank 3) and 39.5 percent of the less worthy volumes (rank 2) had never circulated. It is interesting to note that those titles which were not rec- ommended at all (rank 1) fared better than the more specialized works of rank 3. This may indicate that the discriminating factor for the undergraduate is level of presenta- tion rather than the quality of the book. The difference between rank 1 and rank 3 is not, however, statistically significant. The Spearman rank-order correlation co- efficient between circulation and strength of recommendation is Rs = .137 (p = .008) for 310 titles. When the 14.6 percent of known faculty circulations is removed from the sample, the correlation does not change appreciably. A separation of the humanities titles from the social science titles gives a differ- ent picture. Figures 1 and 2 indicate a stronger relationship between circulation · and evaluation for titles in the social sci- ences and no relationship at all for titles in the humanities. The Spearman rank-order correlation for titles in the social sciences was Rs = .233 (p = .002), but the same test applied to the titles in the humanities was Rs = .043 (p = .3). CONCLUSION The reviews appear helpful in identify- ing the most worthy titles as those most likely to be used repeatedly. No selector would want to ignore recommended titles of which 41 percent are likely to circulate three or more times in two years. Similarly those titles appealing primarily to a more elite audience of specialists ought to be scrutinized if the selector is concerned about maximum use. The question of the level on which the book is written is an im- portant one, to judge from the statistics of 378 College & Research Libraries September 1983 use. Evidently undergraduates can deci- (1')000 ~ c:i...0(1) pher (and reject) a title because of its spe- '* C")C")Cf') cialized appeal more readily than they can 0 !-< determine whether a book is poorly orga- ~ 0 nized or argued. ~ · -.:t<-.:t ~~ ?fl."' tute, whereas the volume on the Olympic § ~~~ ~ . >-< 0 ~~ Games may be only one of many alterna- ~ "'u ~ ~ ~ tives. Circulation is an easy gauge with ~ z ~~ which librarians can take the measu:te of a ~ 0 "' < ~ ~ ~~~;:!: c)...-. collection, but it records only use, not use- f-1 < z-.o fulness. s The librarian selecting strictly on the ba- u ~ c; ~Cf') sis of probable popularity runs the risk of §~ lf)C")N ~ ~~~ ~ . developing a collection which could be 0 ~~ 0 Nu categorized as "lightweight" academi- >-< ~ u cally. Similarly, collections based exclu- z ~ sively on Choice may build a collection of ~ ~ oll)OC") cioo worthy titles which may or may not ad- a z...-....-....-. ZC") dress the needs of a particular institution's § undergraduates. As stated earlier, Choice ~ § -.:t< 00"' ~-.:t< does not recommend this latter strategy, ..2~ or:-:,....; ~ . either. Some local factors which ought to 0 Cf'J~N o"- ,....u ~ influence patterns of collection develop- ~ "' ment could include the size of a depart- ~ ment, class enrollment, frequency of a c)['o.....-.lf) oM z ,..... ZN course offering, term paper assignments, past library use, and the likelihood of cross-disciplinary interest. I=: This study does not dispute the point -~ rG that college librarians may very well want ~=::"3 to acquire those titles garnering critical ac- ~=:: -9.~ claim, regardless of the subsequent circu- '0 .9] ~ '"iii lation record. Nor should librarians feel e-§ ]8!5 0 c ·.= 8·o e they are alone with their worthy, uncircu- ~"' E-< ::S"3 ·o~ ti cru lating volumes. The publishing industry ~-!:l r.t.U o...-.M 50% 40% 30% 20% 10% 50% 40% 30% 20% 10% Research Notes 379 TITLES WITH ZERO CIRCULATION D social Science Sample ~ Humanities Sample 5 4 3 2 REVIEW RANK, HIGHEST TO LOWEST FIGURE 1 TITLES WITH THREE OR MORE CIRCULATIONS D Social Science Sample ~ Humanities Sample 5 4 3 2 REVIEW RANK, HIGHEST TO LOWEST FIGURE 2 380 College & Research Libraries itself expects to lose money on 80 percent of the books it publishes; the problem oc- curs in recognizing which 80 percent. 1° Fi- nally, no one doubts that undergraduates don't read Choice. But as Evan Farber has September 1983 pointed out, perhaps it is our responsibil- ity to further educate library users in the value of knowing ''how to select books be- fore reading them, not just how to use the card catalog. " 11 REFERENCES 1. The Idler, no.86 (1 Dec. 1759). 2. Choice, 1:13-15 (Mar. 1964). 3. Ibid. 4. Beth Macleod, ''Library Journal and Choice: A Review of Reviews,'' Journal of Academic Librarianship, 7:23-28 (Mar . 1981). 5. Daniel Ream," An Evaluation of Four Book Review Journals," RQ, 19:149-53 (Winter 1979). 6. Bowker Annual of Library and Book Trade Information. (26th ed.; New York: Bowker, 1981), p.328, p.355. 7. Richard Hume Werking and Charles M. Getchell, Jr ., "Using Choice as a Mechanism for Allocating Book Funds in an Academic Library," College and Research Libraries, 42:134-38 (Mar. 1981). 8. Allen Kent, and others, Use of Library Materials: The University of Pittsburgh Study. (New York: Dek- ker, 1979), p.10. 9. Beth Macleod, "Library Journal and Choice: A Review of Reviews," p.27. 10. New York Times, (15 Jan. 1982), sec. 3, p.22. 11. American Libraries, 7:515 (Sept . 1976). A Methodology for Estimating the Size of Subject Collections, Using African Studies as an Example Joseph J. Lauer This note provides a formula for esti- mating the number of Africana titles in large libraries using the Library of Con- gress classification schedule. The meth- ods used to establish this formula could be used for other subjects, and an analysis of the completeness of LC class numbers for a given subject would seem to be an essen- tial first step in developing a useful con- spectus or in compiling a questionnaire on subject strengths. Shelflist measurement has become a fairly common method of determining the number of titles held in broad subject cate- gories. But most subjects are scattered to some extent, and this problem is espe- cially severe with area studies. For exam- ple, books about Africa are found throughout the LC schedule, with a con- centration in the DT section. Thus, before one can make an objective estimate of the total number of Africana titles, it is neces- sary to establish what percent of all cata- loged Africana faHs in . the DT section (which is exclusively Africana). There are at least two ways to determine the percentage of all Africana falling in the DT section: (1) describe and analyze the Joseph J. Lauer is Africana bibliographer, University Research Library, University of California, Los Angeles .